OpenAI’s July ups and downs: GPT Health rollout, hot IPO anticipation offset by rogue AI hacking, medical misconduct lawsuit

What a month OpenAI is having. At times it’s been hard to look away for general news consumers as well as the AI-watching press. Here’s a summary scorecard of high points and low points in order from most to least recent.  

UP: OpenAI announces the launch of Health in ChatGPT to all U.S.-based adult users across all plans. 

This means that, if you have ChatGPT in any of its six pricing tiers, you can query it for health info from anywhere inside the app. You used to have to go to a dedicated health hub. 

In a July 23 announcement, OpenAI says its designers noticed that health questions often came up in conversations about other topics. 

“For example, while planning meals or looking for recipes, someone might want ChatGPT to take a food allergy or dietary restriction into account,” the announcement explains.

Among members of a test group, more than 70% of health-related conversations took place outside the dedicated health experience, OpenAI shares. “Moving to a separate space added an unnecessary step,” the company notes. 

The company underscores that all ChatGPT conversations are encrypted at rest and in transit, and information connected in Health “receives additional encryption protections.”

DOWN: A man is suing OpenAI and its CEO Sam Altman for some dangerously bad medical advice he says ChatGPT gave him. 

The plaintiff, one Scott Winters of Florida, says the algorithm talked him into forgoing treatment for symptoms that ended up being a life-threatening pulmonary embolism. 

Winters, a former pastor, filed the suit July 22 in San Francisco County Superior Court. 

Winters argues ChatGPT “crossed the line from providing information into practicing medicine without a license,” numerous outlets are reporting. 

In its defense so far, OpenAI stresses that its chatbots are not meant to replace medical professionals, according to numerous news accounts.

The lawsuit text alleges that ChatGPT-4o “frequently opined on and evaluated Scott's physical condition, conveying the authority and confidence of a medical professional rendering a clinical assessment and treatment recommendation, despite not having the qualifications to do so.”

DOWN: OpenAI’s advanced AI models hacked into the internal systems of Hugging Face, an open-source community for AI enthusiasts. And the models did the dirty deed of their own accord and volition.  

One of the most troubling aspects was the speed with which the models broke in. 

The nonhuman prowlers “spent mere hours carrying out a hack that would have taken a skilled human far longer,” Bloomberg reported July 23. “Typically, even a talented hacker would need a couple of weeks to complete an attack like this.”

For its part, OpenAI says it expects this sort of thing to happen with some regularity as “increasingly cyber-capable” models proliferate. 

“We consider this incident to be an unprecedented cyber incident involving state-of-the-art cyber capabilities and are responding accordingly,” OpenAI assures the general public in a July 21 post. “We will continue to conduct a thorough investigation alongside Hugging Face and will share more details on the vulnerabilities, incident and findings when our investigation is complete.”

HealthExec has more coverage of the incident here. 

IT’S A WASH: OpenAI is facing very difficult market conditions as it approaches its hotly awaited IPO. Then again, balancing this out is the fact that arch-rival Anthropic is in the same boat.

“Whichever company goes first is likely to determine the trajectory of the one that follows,” Forbes senior contributor Peter Cohan observed July 9. 

If Anthropic goes out first and its stock skyrockets after the IPO and keeps rising, the same is likely to happen with OpenAI, Cohan forecasts.

“If the first IPO goes badly, however, the same thing could happen with the second one,” he adds. “Or a bad first IPO could delay further the second one.”

In any case, both AI monsters are “poised for two of the biggest stock market debuts ever—potentially topping $1 trillion each,” the San Francisco Chronicle reports, evidently licking its chops over the potential fiscal windfall for the City by the Bay.